Math 213br HW 12 solutions May 5 2014 Throughout X is a compact Riemann surface. Problem 1 Consider the Fermat quartic defined by X 4 + Y 4 + Z 4 = 0. It can be built from 12 regular Euclidean octagons by gluing them together so that 3 meet at every vertex. In terms of (X Y Z where are the centers of these octagons? Where are their vertices? Put ζ = e 2πi/8. The centers of the octagon correspond to the 12 points that are fixed points under an automorphism of order 8. An automorphism of order 8 is [X : Y : Z] [Y : ix : Z] whose fixed points are [1 : ζ : 0] and [1 : ζ 5 : 0]. The remaining points are obtained by the orbits of these two under permutation action by S 3 on the homogenous coordinates. The vertices correspond to the 32 fixed points of an automorphism of order 3. One such automorphism is [X : Y : Z] [Z : X : Y ] whose fixed points are [1 : ζ 3 : ζ 2 3] where ζ 3 is any third root of unity. The orbit of these three points under the automorphism of order 8 we wrote down above are the 32 points corresponding to the vertices. Problem 2 Let S be the oriented topological surface obtained by identifying opposite edges of a regular 4g-gon P in R 2. Then the edges of P taken in order determine cycles C 1... C 4g in H 1 (S Z. Compute the intersection pairing C i C j and prove directly that the matrix J ij = (C i C j with 1 i j 2g has determinant ±1. We ll do the case g = 2 The generalization to g > 2 is straightforward. We ll label the edges by the patten on the left; then the diagram on the right shows the vertex of the octagon:
Problem 3 2 C 1 C 4 C4 C1 C4 C 1 The intersection matrix can be read off from the diagram at right: 0 1 1 1 I = 1 0 1 1 1 1 0 1 1 1 1 0 The following matrix is the inverse: I 1 = 0 1 1 1 1 0 1 1 1 1 0 1 1 1 1 0 For instance here s how you might convince yourself. The intersection matrix is invariant under rotating the octagon which is equivalent to saying that T I = IT where T is the matrix 1 1 1 1 that corresponds to a rotation f the octagon by 1/8. Now observe that the matrix we called I 1 has the form [v T v T 2 v T 3 v]. Then once you check that Iv = e 1 it follows that IT j v = T j Iv = e j so II 1 really is the identity. Since I is an integral matrix with integral inverse its determinant is ±1. Problem 3 Let X be a compact Riemann surface of genus g. Use the Riemann-Roch theorem to show directly that the map ϕ : X g Pic g (X is surjective. Here Pic g (X is the group of linear equivalence classes of divisors of degree g on X and ϕ(p 1... P g = [ P i ]. When is dim ϕ 1 ( g 1 P i > 0? Math 213br s HW 12
Problem 5 3 Let D be any divisor of degree g. Riemann-Roch gives h 0 (D = 1 g + g + h 0 (K D 1. So we can find a nonzero meromorphic function f such that (f + D. Hence D is linearly equivalent to a sum of g points with multiplicity i.e. D is in the image of X g. The dimension of the fiber over D is just the dimension of the space of effective divisors linearly equivalent to D which is equal to h 0 (D 1. Hence the fiber has positive dimension whenever dim h 0 (D > 1 or equivalently whenever h ( K D > 0. These are by definition the special divisors. Problem 4 Describe the Jacobian of the curve X defined by y 2 = x(x 4 1 as a principally polarized complex torus. That is write Jac(X = /Λ for an explicit lattice Λ and give the symplectic form on Λ = H 1 (X Z. Then describe the action of the automorphism T (x y = (ix ζy on Jac(X where ζ = e 2πi/8. In particular describe the action of T on H 1 (X Z and show it preserves the symplectic form. X is isomorphic to the Riemann surface obtained by gluing the regular octagon as in problem 2 with the automorphism T acting by rotation by one eighth. In problem 2 we wrote down the symplectic form I and the action of T with respect to the basis {C 1 C 4 } of H 1 (X Z. Using that T C i = C i+1 for 1 i 3 and T C 4 = C 1 and the computation of intersection numbers in problem 2 we see that T preserves the intersection form. Choosing the basis for Ω(X to be ω 1 = dx ω y 2 = x dx we see that T ω y 1 = dix = ζy ζω 1 and T ω 2 = ζ 3 ω 2. Now we have ω = T ω. Put T C C C 1 ω 1 = a C 1 ω 2 = b the matrix of periods is given by ( a b bζ 3 2 bζ 6 3 Hence we must have a 0 b 0 and we could normalize ω 1 ω 2 so that a = b = 1. Problem 5 Let G = Z 6 be the subgroup of Div(X generated by the Weierstrass points P 1 P 6 on a Riemann surface of genus 2. 1. Show that 5P 1 6 2 P i is a principal divisor. 2. Find generators for the subgroup of all principal divisors H G. bζ Math 213br s HW 12
Problem 6 4 3. Describe the image of G in Pic(X. 1. One can choose an isomorphism of X with a hyperelliptic curve given by y 2 = f(x where deg f = 5. The Weierstrass points are the ramification points of the double cover x : X P 1. We can choose f so that the zeroes of f correspond to P 2 P 6 and the unique point at infinity is P 1. Then the function y has zeroes at P 2 P 6 and pole at only P 1 hence div(y = 5P 1 6 2 P i. 2. We will show that 2P i 2P j and 5P 1 6 2 P i generate H. We already know that they are principal divisors. Let G 0 denote the subgroup of G consisting of elements of degree 0. The above elements generate a subgroup of index 16 in G 0. We have a homomorphism ϕ : G 0 Jac(X which factors through Jac(X[2] = F 4 2 since 2P i K is canonical. The kernel of ϕ is exactly H. We claim that ϕ is surjective. Indeed it suffices to see that the 16 elements P i P 1 (1 i 6 P i + P j 2P 1 (2 i < j 5 are pairwise not linearly equivalent. Indeed any linear equivalence among those elements will give rise to P i + P j P k + P l which can only happen when when either {i j} = {k l} or i = j and l = k due to the uniqueness of the hyperelliptic involution. Thus ϕ is surjective and H has index 16 in G 0 hence it coincides with the subgroup generated by 2P i 2P j and 5P 1 6 2 P i. 3. By what we have shown above the image of G 0 is Jac(X[2] and thus the image of G in Pic(X is an extension of Z by (Z/2Z 4 so it is (Z/2Z 4 Z. Problem 6 Recall that Ω(X carries a natural inner product α β = (i/2 X α β. 1. Given an orthnormal basis (ω i for Ω(X we define the Bergman metric on X by ρ(z 2 dz 2 = ω i (z 2 dz 2. Prove that ρ is independent of the choice of basis. 2. Prove that we can identify Jac(X with C g /Λ in such a way that the Abel- Jacobi map X Jac(X is an isometry from the Bergman metric on X to the Euclidean metric on C g /Λ. 1. The Bergman metric has the following interpretation: A vector v T Xx is a linear functional on Ω Xx and hence on Ω(X via the evaluation Ω(X Ω Xx. The Bergman metric assigns to v the norm as a functional on the normed space Ω(X. Since this description does not involve choosing orthonormal bases for Ω(X it is obviously invariant under change of such orthonormal basis. Math 213br s HW 12
Problem 6 5 2. Using an orthonormal basis ω i and fixing a base point O we can write the Abel- Jacobi map as AJ Q : P ( P ω O i which identifies Jac = C g /Λ. If v T XP is a tangent vector the above map sends it to the vector (ω i (P (v thus the Bergman norm of v agrees with the standard Euclidean norm on C g thought of as the tangent space of Jac(X at AJ Q (P. This shows that AJ Q is an isometry. Math 213br s HW 12